Forum Moderators: phranque
For instance, my logs seem to suggest that one IP address is used to request the HTML page, CSS and some images, then a second, very close IP address is used to request the remaining images for the page.
Why would someone want to split up their requests across IPs on the same network like this?
Netscreen firewalls (and, presumably other brands, as well) have a feature that will randomly spread NATed traffic over multiple IP addresses on the untrust interface. This is meant as a security measure. (Making the browser less uniquely-identifiable.)
A router or firewall might also be set-up to spread traffic over multiple interfaces. (e.g. multiple internet connections). This would typically be done for load-balancing, though it would also more-effectively accomplish the security goal above.
Similar techniques might also be used to provide greater bandwidth by using multiple low-speed connections where high-speed connections are not available.
They do it to drive log analyzer's batty, I think.
... or it may be a technical side effect of running a loadbalanced farm of parallel proxy servers for their many customers they have. Have seen this with some other large entities, too, where just one single proxy won't be sufficient -- due to traffic volume and/or for failover/redundancy reasons.
Kind regards,
R.